EPCO-20. DYNAMICS OF TUMOR GENOME SHAPING AND CLINICAL RESPONSE IN GLIOBLASTOMAS UNDER IMMUNE CHECKPOINT INHIBITORS

Abstract

Abstract Immune checkpoint inhibitors (ICI) have shown remarkable success in several tumor types; however, this success has not been common in glioblastomas (GBMs). Given the immune-suppressive tumor microenvironment, the timing when to apply ICI may be a determining factor for clinical success. With this hypothesis, we compiled a longitudinal cohort of GBMs treated with different ICI schemes and a control cohort (upfront ICI (n=16), recurrent ICI (n=23), and control cohort by standard of care (n=33)). We investigated the genomic correlates to clinical response. Neither ICI itself nor different timing strategies failed to show a survival benefit for GBMs. However, the upfront-ICI cohort had more long-term survivors (overall survival >= 36 months) than the recurrent-ICI cohort (37.5% (n=6/16) vs. 26.1% (n=6/23)). Genomic analysis revealed that alterations in the PI3K-pathway or TP53 were associated with a poor response to ICI, while BRAF mutation was related to a favorable response. Interestingly, only TP53 alteration was associated with a worse prognosis in GBMs treated with upfront ICI. Tumor mutational burden (TMB) or the amount of neo-antigens failed to show any correlation with the response to ICI. However, ICI-treated GBMs experienced divergent evolution regarding the TMB, while GBMs treated by standard of care only exhibited unidirectional change. We also explored the immunogenomic shifts following different treatment modalities. ICI-treated GBMs showed more variable changes in both HLA alleles and neo-antigen burden (p-value=0.41 & 0.008, respectively). Among the upfront-ICI cohort, GBMs with a better response showed an increased neo-antigen burden upon recurrence (p-value=0.03), suggesting the selection of fitting clones during therapy. Collectively, we observed that that GBMs experienced distinct evolutions under different treatment schemes. The vulnerability of GBMs to ICI should be considered in the context of tumors’ capability to dynamically shape the tumor genome to fit into therapeutic pressure.